[llvm-exegesis] Allow benchmarking arbitrary code snippets.
Summary:
This is a step towards fixing PR38048.
Note that right now the measurements are given per instruction. We'll
need to give measurements a per code snippet and update the analysis (PR38731).
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D52041
llvm-svn: 342947
diff --git a/llvm/tools/llvm-exegesis/lib/Assembler.h b/llvm/tools/llvm-exegesis/lib/Assembler.h
index 76030ae..f2a7716 100644
--- a/llvm/tools/llvm-exegesis/lib/Assembler.h
+++ b/llvm/tools/llvm-exegesis/lib/Assembler.h
@@ -18,6 +18,7 @@
#include <memory>
+#include "BenchmarkCode.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/BitVector.h"
#include "llvm/CodeGen/MachineFunction.h"
@@ -39,12 +40,6 @@
// convention and target machine).
llvm::BitVector getFunctionReservedRegs(const llvm::TargetMachine &TM);
-// A simple object storing the value for a particular register.
-struct RegisterValue {
- unsigned Register;
- llvm::APInt Value;
-};
-
// Creates a temporary `void foo(char*)` function containing the provided
// Instructions. Runs a set of llvm Passes to provide correct prologue and
// epilogue. Once the MachineFunction is ready, it is assembled for TM to